the National Cancer Advisory Board of the National Cancer Institute (NCI), made a statement regarding randomized clinical trials (RCTs) that, although helpful in some circumstances, they were not always possible and that clinical judgment coupled with descriptive and other types of studies were also necessary in making proper recommendations for clinical practice.
In 1948, the first RCT was used in the testing of antituberculosis drugs. In 1957, with the establishment of the NCI, the Cancer Chemotherapy National Service Center, and the End Results Group, RCTs came into vogue for the evaluation of cancer treatments. The RCT was designed to minimize selection bias and is considered the best method of evaluating differences between two or more treatments or no treatment at all.
Descriptive studies were still used for Phase I studies of drug toxicity and Phase II studies determining optimal drug dosage for tumor regression. In the fields of surgery and radiation therapy, descriptive studies were useful in measuring the degree and length of tumor regression, stage specific tumor free intervals, recurrences, morbidity, survival, mortality, and functional outcome. However, RCTs were invaluable in determining the best of several treatments. It is doubtful if breast-conserving surgery plus radiation could ever have been proven as effective as modified radical mastectomy except through the use of an RCT.
In 1962, the first randomized clinical screening trial (RCST) was designed to evaluate whether screening normal individuals could detect cancers earlier, resulting in a decrease in mortality. Although the RCST minimized selection bias, a new set of problems were created that were not encountered in treatment trials.
Trial Size
The trial size suddenly became immense. Due to the low incidence of any one cancer, it became necessary to test 100, 1000, or 10,000 normal individuals to detect a single cancer. A high detection rate was considered to be 10 cancers per 1000 examinations as in the initial screen for breast cancer (a 1% yield). Instead of randomiz-Address for reprints: Charles R. Smart, M.D., ing single patients as in RCTs, in RCSTs it became necessary to 1262 Chander Dr.,